Beruflich Dokumente
Kultur Dokumente
November 1981
Abstract: This report briefly describes robot technology and goes into more depth about
where robots are used, and some of the anticipated social and economic impacts
of their use. A number of short term transitional issues, including problems of
potential displacement, are discussed. The ways in which robots may impact the
economics of batch production are described. A framework for analyzing the
impacts of robotics on ecnomywide economic growth and employment is
presented. Human resource policy issues are discussed. A chronology of
robotics technology is also given.
This research was supported by the Industrial Affiliates Program of the CMU Robotics Institute and by
the Department of Engineering and Public Policy.
A version of this paper will appear in Technology Review in the spring of 1982.
I
I-is-i 0%Figures
Figure 1: Estimates of U.S. Robot Population, 1970-1981 3
Figure 2: Comparison of Production Technologies 11
Figure 3: Disiribiition of Value Added in flit?Engineering 1i:dustries by Batch Size 12
Figure 4: Distribution of Value Added in Mnniifacturiny by Batch aiid Mass Production 13
Figure 5: Robot Serving a Cell 17
Figure 6: Flexible Computerized Manufacturing System 18
Figure 7: Categories of Producer's Durable Equipment 23
Figure 8: Economy Wide Impacts of Improviny ManuFacturing Productivity 24
Figure 9: Sex/Race Distribution of the blanufacturing Workforce, 1980 31
Figure 10: Sex/Race Distribution of Manufacturing Operatives and Laborers 32
Represented b y Labor Organizations
Figure 1 1 : Analyzing Economy Wide Employrwnt Issues 40
Table 1: Robot Capabilities 5
Table 2: Overview of Robotic Sensors 7
Table 3: Classification of Industrial Robot Tasks 9
Table 4: Ratio of Production Workers to Robots, Mid 1981 15
Table 5: Estimates of Productive Cutting Time in Metalworking Manufacturing 21
Table 6: Estimates of Average Machine Tool Utilization in the Metalworking 22
Industries,l977
Table 7: Motivations for Using Robots 26
Tab le 8: Prime Operative Tasks for Level I Robots 29
Table 9: Prime Operative Tasks for Level II Robots 29
Table 10: Annual Average Turnover Rates in ManiJfaCtUring, 1980 33
Table 1 1 : Aye Distribution of the Manufacturing Workforce, 1980 . 34
Table 12: Major Unions Representing Workers in the Metalworking Industries 35
Table 13: Wage and salary Workers Represented by Labor Organizations, May 1980 36
Table 14: Characteristics of Union Clauses Relating to the Introduction of N e w 38
Technology
Table 15: Enrollments and Completions in Public Vocational Education in Selected 44
Metalworking Occupations: National Totals: FY 1978
1 THE IMPACTS OF ROE30TICS
Robots are valued in industry for the usual qualities of machines: untiring availability, predictability,
reliability, precision and (relative) imperviousness to hostile environments. They do not, as yet,
possess several important capabilities which come naturally to humans: the ability to react to
unforeseen circumstances or changing environments, and the ability to improve performance based
on prior experience. State-of-the art robots (mostly in, , research labs) d o have crude senses of "sight"
and "touch", and limited capability to coordinate their manipulators with sensory input. Because of
current limitations, today's robots are usefully employed in highly structured. industrial environments
where practically all of the variability and decision making can be engineered out of the workplace.
Existing uses of industrial robots all involve repetitive preprogrammable tasks such as spot welding,
spray painting, palletizing, and the loading and unloading of many types of metal forming and metal
ctitting machines. The next generation of sensor based robots will be able to perfirm a broader
range of tasks under less structured conditions, in addition to becoming cheaper and easier to use.
Expected uses of robots with vision and improved feedback control will include inspection, asseiiibly,
heat treatment, grinding and buffing, and electroplating.
Eventually, many of the "hands on" tasks performed by production workers on t h e factory floor will
be done by robots in. computer controlled manufacturing systems. Programmable automation is
beginning to replace the current generation of manually controlled machines. This transition will
undoubtedly continue for many decades. There is a potential for significantly improving the
productivity of our manufacturing sector, and increasing the wealth producing potential of the
economy as a whole. We also face significant social impacts , such as the short term prospect of
technological displacement, and the longer term prospects of basic structural shifts in the economy.
The term "robot" allegedly stems from the Czeck word "robotnik", meaning serf. It was first
introduced into the popular language by Caret Capek, a Czeck Playwright, in R.U.R. The concept of
i)rocJr~imiii~il)l:? machiiiery. however, (fates b x k niiich e:trtic?r. to the 18 t h cer,fcir-y. when llie
Frenchman Bouchon. Vacaunson,Basile. Falcon and Jacquard developed mechmical loonis which
were coi?trollt.d by punch cards. Spericer's Automat. a cain prograinnxible lathe used for producing
screv/s. fllits. ;tt1(.1 gt?;irs. V J ~ controlIi'd
S b y f;ttiil<j CJLJidZS to Ill5 cild Of a rotating cli-uin in the inid
1S 70's.Since ttiert. rnecli:iriical controls have proliferntecf iii 11i?iitachii ie tool iiidtistry.
2
~ . . ~ ~ c h ~ nrn?r,ipiilators
icel 3;so h::ve 2 Icng f>iStCr)i. In 1Es2, %:./Xd !b?k;!t, 0: Pitts?ll;(;h, p2!2nfed
a TGiZrY c:aii~ viith a motc;ized s:i;2>: for reniot;ing ir;c_;oisoL;t of ~LI:CXCS. The first jGii-,f-3.d
mechanical Zrrn v;hich coiild c o u ! ~ !>!a;; t ; x k C? szries cf motions $.vas c ' ~ v - s - : o gby
~ ~Poli~i-cJ
J in lcJ3r-j.
The n?achine VJBS specialized for spray painting. The first gznerc?! ,mr?ose p!a;ibac% [Init fc;
contro:!ing machines wr?sdsveloped by G m r g e Dnvcl in 1945. I-le i i c w s e d the &vice to Remington
Rand, \Vho intended to use it for the Univac; Coniputer, :;.hich w3s just devdoped. The ccntrol!a was
not fast enough for the desired purpose, and the paient was returned to Devol. In 1954, DCYJOI
developed !he first general purposc manipu!ator with a playback mtrnory arid soint-io-point can trcl.
The pztent for this P:ogrammed Arlicle .Trmsfer was issued in 1961. The patent'siates, Universal
Automation, or "Unimation", is a term ;hef may well characterize the generel object of this inven:ion. .
Devol's early patents \@re sold to Condec, and formed base for Condec's robot division, Unimaiion,
Inc. In the p t r l o d between 1954 and 1963, Devol and several others patsn:ed the niajor featurss of
the first generation of robots.
Early robots had cornpuier like functions, such as memory, but were mzde up of electronic logic
components "hardwir.ed" to perform a specific set .of tasks. Electronic controls w3re used to
essentially duplicate the functions of othe: "hard automated" control functions. Robots controlled b y
general purpose computers wer8 developed in the early 1970's. The first mini computer controlled
was commercialized in 1974 by Cincinnati Milacron. Microprocessor c o n t r o l l ~ drobots follo~ved
several years later. The computer conirolied, or "soft wired" robot, is far more powerful than a
machine with specialized electronic logic circuts. It can w o r k in severzl coordinatg systems. be
programmed "off-line", interface with sensors, and so on. Computer controll& robots a r t no1.v
becoming very s2ecialized peripheral features of a General purpose. computer. A partial, and still
preliminary chronology of significant developments in robotiCs is given in tha apbendix.
The Japanese have more experience in robot applications, even though robots w8re originally I
developed, and first ap9lied in tha United States. According to Paul Aron (Aron,81), as of the
beginning of 198i; there were over 11,OGO machines in Japan which match the definition of industriai
robots applied in the United Stztes. Comparing this Count to cur US oopulation es!imates, the
Japanzse have nsarly three times as many industrial rOSOtS ir1s:ailed and operating as the United
states.'
3 THE ILfPACTS OF ROBOTICS
5500
5000
4500
4000
3500
3000
2500
2000
1500 . #(partial count)
1000
500 A
Cartesian (rectilinear)
Cylindrical
Polar
Each of these systems has three degrees of freedom, sufficient for the arm to reach any point within a
volume of space defined by the maximum extension of the arm.3 Of these types, the anthropomorphic
revolute (or polar articulated) architecture, requiring only cylindrical couplings, offers komparatively
large working volume with minimal spacial intrusion and good ability to avoid obstacles along the
position path. The chief disadvantage of polar architectures has been that servo controls for
continuous path operation are more sophisticated than controls required for the other architectures.
However, recent advances in computer processing power have effectively eliminated this drawback.
For this reason, Cartesian and cylindrical architectures are likely to assume reduced importance in the
future, except where exceptional positional accuracy is needed.
As three degrees of freedom are required to reach any point within the working volume, three
additional degrees of freedom are required to deliver the tool or workpiece in any arbitrary
orientation. This may not be necessary in some cases, e.g. i f the workpiece is cylindrical or spherical.
Most robots have some type of articulated wrist, giving them the addjtional degrees of freedom as
needed.
The performance characteristics ,of robots without sensory feedback can be summerized under
four headings:
* self diagnostic
* 400 hours for mean fault tracing
Reliability
time between failure
o reliability
0 p ro y rain m a bilit y
More detailed discussions of these characteristics and extensive references are found in
(Toepperwein,l980) and (Engelberger.1980). Temporary limitations of robots relate to the speed of
the arm, the amount of force or payload it can deliver, the precision of the motions, the ease of
programmability and the complexity of sequence of actions it can be instructed to do. There are
significant tradeoffs between the various performance characteristics. Extreme accuracy is available
from robots with only three or four dsgrees of freedon1,a very small payload, and a relatively tiny
working volume. Such robots may be appropriate for limited operations with very small parts, such as
assembling watches or cameras. On the other hand, robots capable of handling large payloads over
siinificant working volumes do not, as a rule, achieve very precise?positional accuracy.
Present manipulators are still far inferior to human arms, and are unsatisfactory for many
applications, due to limitations on speed, accuracy, and versatility. Transmission mechanisms, such
as gear trains, lead screws, steel belts, chains and 'linkages used to transmit power from motors to the
load cor;strain performance capabilities. New robot designs, such as the direct drive manipulator
developed at Carnegie-Mellon (See Asada,81), make it possible to remove all the transmission
mechanisms between motors and the load, and pave the way for a new generation of light weight,
high performance robot arms.
The more fundamental limitation on present day robot capabilities relates to the need for pre-
specification of t h e task in complete detail. Most tasks in the real world cannot be pre-specified to the
required degree, but require adjustments and modifications as the task proceeds. Picking standard
parts from a bin is trivially easy for humans and exceeding difficult for a robot. The same applies to
cutting logs or fitting pieces of cloth together. The robot must sense the appropriate attributes of the
workpieces as the operation proceeds, and make corrective maneuvers as needed. It must be able to
recognize when the workpiece is damaged and shoiild be removed from the line, and recognize when
the desired result has been achieved. These are major challenges to the state of the art.
Capabilities necessary to overcome the difficulties of coping with non-standard orientations and
variable workpiece attitudes can be summerized under two headings:
0 sensing
Robot sensors are divided into three major categories, following (Raibert. 1981),in Table 2. While the
range of possible sensory inputs is quite large, the problem of interpreting the sensory signals by the
robot's controlling "brain" remains as a separate dimension. The transducers respond to external
stimulation. and provide a stream of input data which is transported to the robot's control system via
communication devices. However,the information cannot be used for purposes of decision making
until computational elements filter, enhance. interpret, and make perceptions on the raw data. Very
few sensors have been Lised in industrial applications to date. but industry and research labs are
actively studying new sensing devices and algorithms for interpreting sensed information.
7 TtiE IMPACTS OF R o t i o - r t c s
I-
- t e r n a l Sensin3 Sensors to measure internal variables important to the control of 3 robotic
n--
mechanism, such as the position and velocity of joints in a manipulator or in a
locomotion system, or internal forces, temperatiires and pressures. Ttiore is no
direct interaction between the sensor and the outside environment. Some type of
internal sensing is found in every type of robotic mechanism.
-
Contact Sensing Sensors measuring touch, force, pressure, slip, or any type of tactile or force inpiit
to monitor the interactions between the robot and its environment. Sinal1
deviations in position which are normally hard to measure can result in very large
forces which are easy to measure.
In tactile or touch sensing, switches, piezoelectric devices, pressure sensitive plastics, ancl strain
gauges are used to measure very sinall forces at a number of points on the robot's
end effector. Except for the simplest on-off devices, tactile sensors are n'ot yet
found on commercially available robots.
Forces are sensed by using strain gquges or piezoelectric sensors to measure all forces ancl torques
transmitted from the robot's end effector to the rest of the manipulator. Forces
can also be measured-at the actuators.
Ranqe Sensinq Sensors.which measure the interactions of the robot and its environment without
any form of mechanical contact. Vision, laser ranging, proximity sensing, sonar,
and radar sense the environment by collecting 2nd measuring reflected energy. In
computer vision systems, TV cameras are interfaced to computer systems to
analyze what is seen. and to act upon this information. Proximity sensors radiate
light over small distances and mdasure the reflected light from a specific volume.
Laser rangefinding is used to analyze a three diniensional geometry. A steerable
laser transmits a spot of light toward the region of interest. The time-of-flight
devices measure the time it takes for the spot to return to determine the distance
to the reflecting object. Triangulation devices displace the receiver from the
source so that the horizontal location OF the reflected spot indicates its distances.
Adapted from:
Marc Raibert, Robotics in Principle and Practice-A Tutorial,
The Robotics Institute, Carnegie Mellon University, 1981.
8 THE IMPACTS OF ROBOTICS
Computer vision has received the most research attention to date of all the range sensing
techniques. Vision systems which could determine the range and shape of an object using the
"structured light" technique were first developed in 1971. In this approach to robotic vision, light is
projected onto the object in a controlled manner. The range is determined by triangulation, and the
shape is inferred form the intersection of the object and the beam. There are several cornrnercially
available sys!ems using the structured light technique. These systems are used to inspect, count,
locate, and orient parts, as well as to guide (servo) a manipulator to an object in real time. More
advanced vision systems which have the capability to use grey scales , stereo ranging and three
dimensional modelling, and which can be programmed to recognize shapes, are approaching
commercialization.
Learning capabilities relate to the creation and modification of an instructional program on-line,
based on a goal statement and sensory input data. Researchers recognize the need for a software
interface to achieve "learning by experience", and high level planning. It is very easy to tell people
what to do, and have them figure out how to do it. Given the instruction, " Put the nut on the screw",
any normal child could accomplish the task without further detail. But today's robot would require
ever each and every detail to be specified in great detail, from how to hold the screw and the nut, to
finding collision free paths. Robot .programming languages can , to varying degrees, plan simple
tasks given instructions. These programming languages are classified in terms of the amount of
knowledge and reasoning power they require of the robot. Explicitly- programmed languages require
the user to specify manipulator positions and trajectories. World-modeNing languages use very
simple instructions merely to specify what is to happen. Manipulator positions and trajectories are
generated automatically.
Clearly. robot programming languages can only be used with 'robots that are controlled by a
general purpose, programmable computer. As of today, very few of the robots currently installed in
the US. and throughout the world are actually computer controlled.
A convenient classification of factory tasks robots are capable of doing, , following (DeGregorio,
7980), is given in Table 3. Robots have initially had the greatest success to date in spot welding
applications, followed the loading and unloading of machine tools, forges, die casting machines and
stamping presses, as well as spray painting, palletizing. and heat treating. Even in these established
applications areas, many practical problems remain to be solved.
Metal cutting machine tools can be loaded and unloaded by hand,. by robots or by integrated
devices fed by automatic transfer lines, as in automobile engine plants. The role of robots here will b e
limited to cases where automatic transfer lines are inappropriate, because a variety of different parts
must be processed, but batch sizes are large enough to justify numerically controlled machine tools,
fed by robots. Because commercially robots cannot yet handle nonoriented parts, the most
successful present applic: 3n is one where the robot unloads one machine and transfers the part to
mother rriachine. The operational linkages between robots and other machines is discussed later on.
A vital task that has attracted much research attention is parts assembly. With minor exceptions,
9 THE II’APACTS OF ROBOTICS
a. Loading/Unloading of Machines:
vii. furnaces
b. Parts hlanipulation
i. packing
ii.sorting
iii. conveying
iv. orienting
c. Palletizing
2. DISPLACEMENTAND PROCESSING
a. Spot Welding
b. Continous Welding
d. Spray Painting
e. Cabling
f. Cutting
a. Dimensional Control
existing asseriibly line jobs cannot be efficiently accomplished by present-day robots for several
reasons, incluc!ing inability to recognize and pick up a desired part from a mixed collection, lack of a
sufficiently flexible multi-purpose gripper, and ,the lack of high level programming languages to
reduce time consuming and expensive setxp procedures. These liinitations can be removed, to some
extent. in newly designed plants where all parts are palletized, or otherwise pre-oriented as they entef
from the outside, and handled automatically thereafter. The other, and more general approach to the
problem is to develop robots with vision,and tactile feedback, or other forms of contact or range
sensing, and that can be programmed "off-line", using high level languages. Another factor which
has emerged through research is that assenibly tasks often must be restructured to exploit the
capabilities of the robot.
In batch production, operations are done repetitively, but only for periods of hours or days, or
maybe weeks. There is a need to perform efficiently, since a sizable number of copies of each
product are made. There is also a need for flexibility, since the machine must be reconfigured for
another product at the end of the run. Only "flexible" types of automation- multipurpose, computer
controlled machines which are easily reprogrammed-- such as robois and- numerically controlled
machine tools are suitable for, batch production.
Robots are not yet cost effective in most custom applications because in such cases, a large
fraction of the labor time is spent settinq un the machines. This still requires the active involvement of
a skilled machinist. Also, prograinming time would typically exceed operation time. For one-of-akind
and prototype products, it is usually easier for a skilled machinist to make the piece then to figure out
how to do it again with a robot. However, developments in computer aided design, such as the
4There is no prwise w a y for identifying industries as batch or mass producers. These estimates are hased on our own
judgement and e ~ p fterice ~t The dillererice between batch and mass production is growing less distinct. and. as as machiney
becoives inoit- I l ~ x i b l ethe
. distinction will become indistinguishably blurry.
11 THE IMPACTS OF F3OHOTlCS
I------
AVERAGE BATCH SIZE .
-
General e EQUIPMENT Special
-e CURRENT
P
V
DOMAIN OF
INDUSTRIAL ROBOTICS
< w
R >
ADAPTED FROM: Mikell Groover.
Automation.Productton Systems, 2nd Computer Aided Manufacturing
I ESTIMATE^
0%
100% I 11.1
-
Relative
Use of L
Source:
Automated Machine-Tool Technology
Parts - American Machinist, October, 1980
Abstracted from the Report of the
Handling - Machine Tool Task Force
-L,
Batch Mass Batch .Mass
(40.3%) (59.7%) (55..3%) (44.7Yo)
9
Metalwo k i n g Other
03her
k!la c h i n e r y Heavy Batch
(2.6%) Machinery Produced
(6.6%) Du rables
(90.8%)
aLitomatic generation of parts programs froin design drawings, are making robotics more applicable
in small batch znd custom operations.
Robots are not generally cost effective in most mass production applications, either, because
specialized mass production machinery can usually perform the operations more efficiently. Cycle
times for today's robots are cornpnrable to h i m a n cycle times, malting it difficult from thein to be used
in high speed work. Mass production machinery, or hard automation, on the other hand, is highly
specialized to repeat a fixed sequence of operations at high sp.eeds for very long periods of time. Auto
engines and transmissions are manufactured in this way. However,it always difficult and expensive--if
not impossible-- to reconfigure the hard automated system for another product. It is usually cheaper
to scrap the machinery, and rebuild the system from scratch. As cycle times are reduced, and
systems designs improve, rob6ts will become more widely used in high speed, large volume
operations.
The important characteristics of the specialized "hard automated " transfer lines used to produce
automobile engines are described in (Taylor, 1979):
...The system is based on a large volume of repetitive but complex machining
operations. Because of precision tolerance requirements in addition to volume production,
large manufacturing capital cost are involved. Except over a very limited range, little
flexibility is inherent in the system to accommodate change. Only a single product is made
with very limited or minor variations, but under a manufacturing environment that is
enginesred to turn out the product in large quantities at minimum cost.
The last sentence reveals the inherent limitations of "hard automation" technolpgy. It is the cheapest
method of production precisely because each element in the system is dedicated to a single function,
for which it is optimized. But the entire plant is virtually a single specialized machine capable of
producing only a single product. Hard automation is also very expensive to install because each
application is custom-made and ,therefore, quite labor intensive.
Most of the batch production industries, and potential robot users, fall within a group of industries
that are commonly referred to as the metalworking sector. This sector includes the following
industries (followed by their Standard Industrial Classification Code)?
Accordiiiy to The 12 th American Machinist inventory of machine tools in the metalworking sector
taken betweal 1976 and 1978, less then three percent of the three million machines in these
15 THE IMPACTS OF ROBOTICS
industries were numerically controlled. According to estimates by several robot inanufacturers, there
are roughly 5000 robots operating throughout all of U.S industry as of mid 1981. of which around &O%
are in the metalworking sector. This means there is roughly one robot for every 1300 production
workers in the metalworking industries, or even more surprising, less then one robot for every 3000
production workers tl~rour~hoiit
all manufacturing.
Sources :
Robot P o p u l a t i o n : C t N R o b o t i c s Survey, A p r i l , 1981.
Employment: Employment and E a r n i n g s , March. 1981: Tab1e.B-2.
Bureau o f . .Labor S t a t i s t i c s .
We see that despite the improvements in computer controlled machine tools and robots over the past
20 years, the production technology in most batch production factories, and in practically all job
shops is still labor intensive and manually controlled . Thus, a large share of all manufacturing is
performed with labor intensive methods involving manual control.
It is no wonder that the United States industry is having problems controlling cost, maintaining high
standards of product quality and improving productivity. Batch production makes it difficult to
optimize machine tool and/or labor utilization. The greater the variability of the product mix, the
harder it is to control the cost and quality standards for a particular product. From a producers point
of view, a variable product mix, and the capability to manufacture new products is highly desirable.
On the other hand, to improve productivity the flows of inputs and outputs must be more tightly (but
flexibly) coordinated and controlled. One of the primary reasons for performance problems in the
U.S.rnanufacturing sector is international competitions is forcing producers to simultaneously
increase both product variety and groduct quality. These simultaneous but mutually interfering
requirernents are pushing existing production technologies and management techniques beyond
t heir CLI r rent capabilities.
npplic;ttion is conimonly seen in several ai'cas such as spot welding. A second type of flexibility
involves shifting an idle-robotto an entirely new task. Our interviews with industrir~lusers suggest that
a pnrticular robot is most often specialized to a particular application, partially due to mobility
constraints. Even though programmable machines are not, as yet, fully exploited for their full range of
flexibility, it is widely acknowledged in engineering circles that flexible automation--or flexible
computerized manufacturing systems (FCMS) is the "wave of the future" for batch production.
The application of industrial robots in activities relating to metal machining cells is receiving
considerable attention. In the next few years, we can expect to see industrial robots being iiistalled in
many medium batch size manufacturing plants, servicing two or three computer numerically
controlled (CNC) machines. There will be a strong emphasis on the use of inexpensive
microprocessors that will coordinate the various pieces of hardware in such a'cell. ' Machine tool
builders are already committed to a strategy in which considerable programmability is enibedded in
the machine tool system itself. Systems are now commercially available that integrate all design and
production stages between generating design drawings to generating the cutting instructions for a
computer numerically controlled cutting tool. Stand alone robots are still crucial to the success of the
total manufacturing operation. Consider the role of the robot in the cell in Figure 5. The part has to be
moved from one machine to another. In addition to such manipulation within the cell, there is a
potential need For robots to carry out preprocessing functions, such as cutting raw bar stock, and
palletizing. There is also a need for supplementary functions, such as deburring. heat treating,
surface plating, and assembly. From a human worker's viewpoint, there are many task within these
activities, such as loading and unloading conveyors or pallets, that are monotonous ahd which suit
t h e capability of a robot. Technical developments that enable robots to be iiiore versatile will clearly
lend to inore widespread installation in the manufacturing industry. For example, the development of
a "universal grippt," or the ability to identify and pick-up a part placed randornly on a moving
conveyor or in a bin are important areas of current research.
In order to carry out e "closed loop machining operation" where the robot may also replace the
routine metrology (measurement and'inspection) operations in a manufacturing cell, tactile feedback
is essential. While some dimensional measurement checks can be made on the machine tool itself
with sensors placed in the tool changer, there will still be the need for measurement off line. Such
measurements are normally done at present by human operators. In moving towards fully automated
cells, the robot will also have to participate in this task via exact placement of parts in measuring
stations. The rapidity with which robots that can "see" and "touch" are developed and accepted is of
particular interest, since these capabilities appear to be vital for applications of robots to assembly
w d inspection.
The next step in systems integration is for several machining cells to be linked together in a Flexible
Computerized hlnnufacturing System (FCMS). Workpieces with a given range of variability can be
cuitoiwtically subjected to diff3ring production processes. as necessary. by niciii2s of very
so[)ljisticated control and transport systems. The developinent of mini and micro computers for
control I~fsrnncle FCMS practical. Robots interact with numerically controlled machine fools and
other ec{uipnient, controlling the sequence of operations. It is more integrated and inore automated
t1l:in ;I traditional "job shop" consisting of mocfitnes in isol:itioii. opt'ratecl by Incliviclual 1iuin;ins. A
niiii\ber of sticli sy:;twi shave been report& in the 1iter;iture. incliidiiicj the Inyersoll- Rand system iii
'
t h - , 1I.S..tlir K c i s ! (;.:;Iiiiaii"Prisina ?" System :ind ill? J:ip:ine:,? syslein known ;IS tile l.l-.:licxlol~-rgy for
17 TI-IE IMPACTS OF ROBOTICS
HI LUNG MACHINE
L1
LATHE
LJ
ROBOT
v \
\
\
r
H
U
P
I
STOCK AND
1
A L E T I Z I N G STATION
MACHIKING CENTER
I 1
18 THE IMPACTS OF ROOOrics
U
19 THE IMPACTS OF ROBOTICS
There is a fundamental reason why robot integrated FCMS may encroach in the traditional mass
production area. It is because of growing coiisumer demand for product diversification, spurred by a
variety of factors, including the introduction of new goods, shorter product life cycles, shifts in
preferences, and a growing desire, and sometimes need, for more customized products. To achieve
true diversity of products a more flexible manufacturing technology will be needed . Production runs
will be shorter and changeovers more frequent. Most important, the need for extensive retooling to
accommodate production redesign must be reduced or eliminated. Curiously enough, the way to
increase flexibility in the mass production of consumer goods seems to be increased standardization
of capital goods. Machines used to mass produce products , such as high speed transfer lines, are
custom built for a single product, or for a small number of variants. As a result, mass-produced goods .
are not as cheap as they could be because they depend on specialized machines and equipment that
are very costly by virtue of being ’custom’ made in very small numbers. “Mass” production would be
cheaper, clearly, if the production equipment itself were also mass produced-or at least produced on
a larger scale. The virtue of programmable, general purpose robots is precisely that a standardized
unit may be utilized in a large number of of different configurations, and situations, achieving
specialization b y software, rather than hardware.
Machines currently used for batch production, such as manually controlled, general purpose
machine tools. or stand-alone NC machines, can be produced in much higher volumes then mass
production machines since one type of machine can be used for a wide variety of purposes. However,
the drawback to the current generation of general purpose, or so called flexible machinery, is that unit
operating cost are high because of low output levels and high labor intensity.6 The development of
high performance, general purpose robots, and their integration into FCMS will eventually permit Lis
to use mass produced machines to mass produce consumer products-a fairly revolutionary change.
The coming revolution in manufacturing technology, among other things, may greatly increase the
efficiency of utilization of machine tools used in batch production. There is an importait implication.
Capital goods-- producer's durable goods including machine tools listed in Figure 7 -- are almost
entirely batch produced. The use of robots and computer control mean that new capital goods will be
much more productive than the old equipment it replaces. If the real cost of manufacturing prodcrcers
durable equipment were reduced as a result of productivity improvements, the price of capital goods
in relation to final products could be expected to decline fairly sharply over the next half ~ e n t u r y . It
~
is difficult to overstate the significance of this event. There would undoubtedly be a ripple effect on
prices of manufactured goods throughout the economy, as outlined in Figure 8. We expect
reductions in the real price of producer's durable equipment to reduce real unit capital cost in the
sectors purchasing this equipment. We expect this effect to, in turn, reduce the real price of final
output of mass produced consumer goods, as well as the real price of output of the
nonrnanufacturing sectors. Final demand might be stimulated to to. some undetermined extenL8
Lower real cost might incidentally have a very beneficial impact on the rate bf inflation. If inflation is
caused by "too much moneychasing to few goods", a sharp increment in productivity is perllaps the
best way to break out of the vicious cycle. These second order effects, while less immediate, may
have greater ultimate importance then the expected direct improvements in labor productivity in
manufacturing.
7Tt,e abSoliJte price of capha1 may not decline. but we expect the price per unit of capability, or quality. to steadily decrease,
as 11.7s beeu the case with computing equipment.
81t IS possible that the real cost of manufactured goods could be reduced without necessarily increasing either real
tll:;pcs.lbk inconies 01 demand. For example. if markets for many categories of standardized goods were nearly satiir3ted.
c o ~ ~ ~ u m : :would
is primalily buy to replace old. or worn oul items. and not to increase their "stock." In this mole wealthier
soci~?t~ p 'r.o l ~ l ecoiild cfioose to increase their leisore time. raltier than increase their real buying power.
21 THE IMPACTS OF Rmo’rics
Productive 6% 8% 27 %
Cutting Time
Reasons for
Lost Time
Load/Unload, 4% 14%
Noncut ting
Set-up,Gauging 12% 7%
Tool Change 7% 7%
Equipment Failure 6% 7%
Inadequate 7%
Storage
Idle Time 2%
Heating equipment
Fa b ric ated
Metal p rod ucts Structual metal products
Forgings and stampings
Furniture
Figure 7: Categories of Producer's Durable Eqiripment
24 THE IhlPACTS OF R(3BO'TICS
I
Improvements in
Manufacturing Productivity
I
I
Y
D u m b e Goods
P ro d 11c e r s
Nan Durable Goods
Producers
I I
-
Transportaion Trade
tL4 i nric
ing'Itu re Corn m u n !cations Finance
Utilities Real Estate
Construction Services
Figure 8: Economy Wide Impacts of Improving
Manii f act i i r inlj Productivity
25 THE IMPACTS OF ROBOTICS
9 S o c i e t a l Benefits Beyond P r o d u c t i v i t y
There arc other major benefits to be gained from robotics, of sca.cely less social significance in the
long run. Thz first of th-zse is to improve the quality of work-life. This is certainly a social benefit, even
though it admittedly Iias a negative side. Throughout history, and continuing today, society has
functioned, in part, by forcing very large numbers of people to perform dull, dirty, dangerous,
degrading and/or demeaning (but necessary) tasks. Machines have gradually eliminated many of the
worst of these tasks over the past two centuries. For example, in industrialized societies, humans n o
longer chop wood, plant, cultivate, or harvest crops by hand. Men no longer carry heavy l od- ds on
their backs. Women no longer have to weave cloth or wash cloths by Iiand. But traditional factories
still use humans for many repetitive materials handling, machine loading/unloacling, tool operating
and parts assembly tasks.
These tasks, in general, make use of the high grade motor skills and natural eye-hand coordination
of humans, without requiring either intelligence, judgment, or creativity. Being repetitive, they are
inevitably boring. To the extend that such tasks* . involve manipulating heavy workpieces, high
temperatures, the use of high speed tools or reactive chemicals, there is also inherently some degree
of hazard. In the long run it can only be counted as a.societal benefit i f such tasks are taken over by
machines, notwithstanding the fact that such tasks currently provide employment and wages for a
number of unskilled and semi-skilled people who are unprepared by education or training to
undertake more demanding kinds of work. Transitional issues and social cost are discussed later.
Survey respondents overwhelmingly ranked efforts to reduce labor cost as their main motivation .9
Users frequently pointed out that the return on investment (ROI) calculation would not be favorable
unless there is a dramatic decrease in direct labor cost. Argumentsfor the benefits of expanding
capabilities, such as improving product quality or increasing production flexibility were often
considered ''nebulous'' b y the financial analyst.
The question was raised as to whether experiwced users learn how to quantify "indirect" benefits
as they accurnulate experience using robots. An executiv? at one firm speculated that inexperienced
users only take direct labor cost into account b?caus? they do not know what other categories of cost
will be affected. He said that his firm had learn:?d how to qiiantify other indirect benefits such as
improved prodirct quality rind reductions in inc?irec.t m:iteri,?l reqrrir-ernerits.Other experienced LiSerS
did not report this kind of "learning".
'f)i;ipei Laboratories. Cainbridge.Mass , c a r r e d out a suivey r a d Ing mOtiv3tIOnS for using assrmbly robots in 1980. Their
rt'spoiidents also ranked direct labor cos! as the priiiiary motivation.
26 THE IMPACTS OF ROUOTICS
'
e To keep up with the Japanese..
Broader strategic concerns such as long term competitiveness apparently are considered, yet they
are seldomly mentioned as the most important motivations. Only one firm said outright that they had
invested heavily in robotics to improve the quality and the competitive standing of their product. They
were also the only firm to give strong emphasis to other "intangibles" such as improved production
flexibility. Interesting enough, this spokesman was the only person among the inany interviewed to
say that applications were not evaluated primarily on the basis of ROI or payback period.
Future uses of robots are not limited to "operative" tasks in manufacturing. On the contrary, some
of the most significant future uses of robots may be to provide feasible means of providing services or
exploiting resources that cannot be provided or exploited at all at present. Handling dangerous
radioactive wastes on a routine basis in a future disposal facility is one example." The choice is
bE?t:Ve?n one kind of mechanization and another: human workers cannot be routinely exposed to
these wastes. Mobile robots would offer a much greater degree of flexibility then telaoperators, or
*' hard aut om at ion.
"
Exploration, mining. construction or other ro:itine activities in hazardous environments are other
examples. Such tasks are C'ifficiilt, dangerous. ~ I I K consizqclently
! inortiirintely expensive. Cotiots rnay
27 THE IMPACTS OF ROBOTICS
find use in coal or other mines, simply because mines are such unpleasant and dangerous work
environments for humans. Robots could drastically alter the economics of conlinercia1 utilization of
space, for example. In the long run, it is likely that if man succeeds in "industrializing" the inoon,
orbiting space colonies, asteroids or other planets, it will only b e done with major assistance from
robots. l'he Viking 2 Lander which touched down on the surface of Mars in September, 1976, is
perhaps only the first of a line of "exploration" robots. Planned Mars surface rover missions will last
8-10 times longer than Viking and entail much greater complexity. The U.S. Navy and a number of
other organizations are actively developing underwater robots or "unmanned submersibles" both for
military 2nd nonmilitary purposes.
Finally, prosthetic robots and household robots exemplify service categories that are increasingly
needed and difficult to obtain in any other way. Paraplegics, and especially quadriplegics, for
instance, might be served full time by voice-activated robots capable of doing a variety of necessary
tasks from feeding to page-turning. Such robots are being developed in Japan. In the U.S., the
Veterans Administration has an ongoing program in Rehabilitative Robotics. The all porpose
hoiisehold "droid" robot is probably a rather visionary idea, at present, but robots could certainly be
designed to perform some types of jobs, notably heavy cleaning. Joseph Engelberger, President of
Unimation, has promised that he wiil soon have a robot (to be named Isaac,after Asiniov) that will
serve coffee in his office. Quasar Industries of Rutherford N.J. built and photographed a model
"household" android in 1978, and announced their optimistic intentions for "mass prodiiction within
two years." The project was somewhat of a hoax, but there is still unquestionable commercial interest
in developing such a product i f only because of the vast potential market. In fact, Nieman-Marcus
Department Stores advertised a household robot (actually a remote controlled device) in their 1981
catelog. For every conceivable application 0 ; an industrial robot. there are at least ten applications
for a household robot. It is impossible to believe that such a vast market will not be exploited at the
earlest possible time.
It is vitally important to recognize the potential importance of some of these applications-- and
some of their adverse consequences--in the picture as a whole. It is entirely conceivable , for
instance,that a century hence historians looking back might say, in effect,"the real significance of
robotics development in the 7980's and 1990's is that they enabled mankind to expand his abode
permanently beyond the earth's surface. and thereby escape the trap of limiied resources associated
with that constraint." All of future history could be very different, depending on whether space is
successfully "colonized" in the next century or not. On the other hand, discounted present value
criteria might tend to put mme weight on proven short-run applications that pay off because of
displaced labor then on very large but very remote benefitslt is to important to assess short-term
benefits and costs, without unduly discounting long-term implications.
28 THE IMPACTS OF ROBOTICS
occupational titles were singled out as having a high potential for robotization, as shown in Tables 8
and 9. The responses to the survey were quit2 varied, reflecting the different requirements of similar
jobs in various industries. The response from each firm depended on its products, the length of the
typical production run, and on the experience of management with robots. Despite obvious
limitations on the completeness of the survey, several occupational categories can still be targeted as
prime candidates for replacement by Level I and Level I1 robots, even though there are some specific
tasks within these occupations that will not be automated for many years to come.
Almost all of the present membership of the RIA--and90 06 of current robot users-.fall within the
metalworking sector. There are nearly three million workers employed in the nine occupations
designated as the prime operative task for Level I and Level I1 robots in the metalworking industries
(SIC 34-37) nationwide. Based on the average weighted response of the percent of jobs which robots
could do, it appears that nearly half a million of these operatives could potentially be replaced by
Level I robots. The figure roughly doubles to one million operatives if Level I1 robots with rudimentary
sensing capabilities were available. Extrapolating the data for metalworking to similar task in other
manufacturing sectors, it appears that Level I robots could eventually replace about one million
operativeis, and Level II robots could eventually replace three million out of a current total of 8 million
operatives. We think the time frame for this displacement is at least twenty years, however.
By 2025, it is conceivable that more sophisticated robots will replace almost all operative jobs in
manufacturing ( about 8 % of todays workforce), as well as a number of routine non-manufacturing
jobs. As we currently understand the situation, concerted efforts should be made b y the private and
public sector to redirect the f u t u r e workforce in response to these changes. Even though several
million operative jobs in the current manufacturing workforce are indeed vulnerable to robotization,
the transition seems hardly catastrophic on a national scale, provided new job entrants are properly
trained, and directed. In our view, the oncoming transition will probably be less dramatic than the
impact of office automation over the same period. By 2025, most current operatives would have
retired or left their jobs . The jobs would not disappear all at once, and robot manufacturing,
programming, and maintenance itself will provide some new jobs. although we think most new jobs
will not be in manufacturing, despite the rapid growth of the robotics industry itself. New "growth"
sectors in the economy, including undersea and space exploration may also provide many new jobs.
The important conclusion is that young people seeking jobs in the near future will have to learn
niarketahle skills other then welding, machining, and other operative tasks fhat are being robotized.
Even though the adjustmmt problems seem manageable, the potential for social unrest in specific
locations cannot be dismissed quite so lightly. Over half of all the unskilled and semi-skilled
"operative" workers--the types of jobs which could be replaced by robots-- are concentrated in the
four major metalworking sectors (SIC 34-37). Almost one half of all production workers in these four
industries are geographically concentrated in the five Great Lakes States-Indiana, Illinois,
tdichigan,Ohio and Wisconsin-- plus New York and California. Within these same states, the
metalworking sector also accounts for a large percentage of the total statewide employment in
mariufacturing. Adjustments in response t0 the rapid diffusion of robotics may be intensified in these
areas. The impacts of not improving the productivity and competitive standing of these very Same
29 THE IMPACTS OF ROROTICS
Inspector 5- 25 % 13 % 5 - 60 % 35 %
Based on 16 responses.
" h l : i s h i w tool operators includes the sepalate types of machinest listed below. These estimates are used as an average to
appi o x i i w a t r ~:ne p.:'centaye of all c3tegoi ies of innchwst listed below which could b e robotized.
30 THE IMPACTS OF ROBOTICS
iridirstries LVIIIalso be concentrated in the same few states. There may also be a disproportionate
iinpact on racial minorities and women. Non whites account for only 11 percent of the natiorlal
workforce, but comprise k t w e e n 15 to 20 percent of of operatives and laborers. (See Figure 9.)
Woinen employed in semi skilled and unskilled manufacturing jobs are less likely to be represented by
labor organizations than their male counterparts. (See Figure 10.) DeFacto economic discrimination
co uId accord inCJ I y increase.
It is often noted that technological displacement would be minimized if the rate of robot
introduction were paced by the attrition rate. At this time, we cannot say whether or not this is a
feasible strategy. An examination of industry attrition rates and of the age distribution of
manufacturing operatives and laborers suggest this strategy is not feasible. According to Bureau of
Labor Statistics data, only one to three percent of the workforce in metalworking [SIC 33-38] leave
their place of work as a result of quits, discharges, permanent disability, death, retirement, and
transfers to other companies. However, these figures may substantially underestimate the percentage
of people transferring out of specific jobs , since they only include people who actually leave the
establishment. Workers who transfer jobs within the same establishment would not be counted in
currently published turnover rates.’*
Contrary to the notion that many manufacturing workers are old and nearing retirement, the vast
majority of the manufacturing workforce still has 20 or more years of active worklife ahead of them.
As of 1980. between two thirds and three fourths of operatives and laborers were less then 45 yt?ars
old, which means that barely a third of the workforce would be retired in the normal w i y b y the year
2000. (See Table 11.) On the average, skilled workers are older, but they are not as likely to be
replaced by robots in the near future.
There are no reliable statistics which cross classify union membership b y manufacturing industry,
hiit it appears that almost all of the membership of the UAW, the IAM. the IUE, the UE, and the USW
work in SIC 33-38, whereas most of the membership of the IBEW works outside of manufacturing.’“
31 THE IMPACTS OF ROt30'lICS
SEX /RACE DI STRI BUT1 Oi'd OF THE MANU F A CTU RING WORKFORCE, 1 980
Percentage distributions:
M: m a l e
F: female
W: white M
NW: non white F
Totals! 88.d 11.4 100.0
SKILLED WORKERS
Othe; Metalworking
Machine Jobsetters: 658000
Iy1-m 7Tot a I s
-
F
Totals 91. 8.q 1.00.0 Totals 92.
SEMI SKILLED AND UNSKILLED 'JVORKERS
Moter Vehic le Other Durable Goods
Equipment 0 eratives: 431,000 Mfg. Operatives: 4,166,000
W NW Totals
M 354 6-c 41.6 M 68.q 16.q 84.3
F 13.q 2.71 15.7
F: female W NW I T o t g k = percent of
LV: white M . 30.3 36-ti 31 -0 occupation group
NW: n o n w h i t e F 17.q 27.p 1 8-9 represented by
(*): Base less Totals 24.q 32.1 25.7 labor organization
F
Totals 56. 56.6
SEMI S K I L L E D AND U N S K I L L E D W O R K E R S
Motor Vehicle Other Durable Goods -
Mf . 0 eratives: 1 917.000
M
F 33. 35.2
Totals 85. 87. 85.8 Totals 46.1 50. 46.8
Totals
SC)lIHCF:
M
F
rrm 41.
32.1
40.1
3L2
40.8 Totals
M
F
(*)
Totals
48.9 54.1
-
41.7
49.2 52.2
ELS,Sept. 1981
l3 Jotzl separations are ter rninations of employment initiated by either employer or. employee. (Rates per 100
employers )Lnyolls are suspensjons wlrhokrt pay for more than 7 consecutive initiated by einployer.( Tordl separdions - Layolls)
inrlci(les qwl:;. discliargf:s.permenjrlt dts3hrlities. retirements .transfers to other establtshmerits. and entrances into the Armed
Forces. Worktrs who change jobs. but do riot leave their place of work are not included in these figures.
34 THE Ih4PACTS OF ROBOTICS
Mac hi ne 658 3.3 15.2 27.8 19.6 17.6 9.4 5.8 1.3 66.0
Jobsetters
Other 638 2.0 9.6 28.8 20.8 20.0 11.o 6.7 1.1 61 .O
Metalworking
Craft Workers
Motor Vehicle 431 2.1 11.3 30.6 25.9 20.1 5.8 3.9 .2 70.0
Equipment
Opeiatives
Other Durable 4,166 5.5 17.5 27.9 19.5 16.6 7.4 4.5 1.1 70.0
Goods Operatives
Non Durable 3,290 6.5 16.0 26.3 19.3 18.5 7.5 4.5 .1.4 68.0
Goods Operatives
Manufacturing 961 9.7 20.3 28.2 16.5 14.1 5.9 3.8 1,.4 75.0
Laborers
Source f o r membership f i g u r e s :
1978:Directory o f N a t i o n a l U n i o n s and Employee A s s o c i a . t i o n s , 1979.
n u r e a u o f L a b o r S t a t i s t i c s , S e p t . 1 9 8 0 . B u l l e t i n 2079
1 9 8 0 : P r i n c i p a l U.S. L a b o r O r g a n i z a t i o n s , 1980. B u r e a u o f L a b o r S t a t i s t i c s .
Collective bargaining contracts are the formal mechanism that unions use to affect company
policies . Union contracts are marked by their large number, and by their diversity of provisions and
their sphere of influence. A comprehensive review of union contracts is beyond our scope. However,
as part of the project on The Impacts of Robotics on fhe Workforce and Workplace, we reviewed
representative contracts from the UAW, the IAM, the IBEW, and the IUE, and identified clauses
relating to the introduction of new technology. The union contracts include clauses relating to job
security, job integrity in the workplace, and benefits to the workers in the event of a lay-off. Job
security attempts to provide workers with guaranteed of continued employment at agreed upon wage
and benefit levels while job integrity deals with the maintenance of the bargaining unit in the face of
changes in the production process. Such clauses encompass concerns relevant to the actual working
conditions of the firm. In the event that job security is not attainable, the unions attempt to ease the
situation of the individLial worker in the period after displacement.
Three of the four union contracts studied had provisions which set up joint union-management
committees to disctiss the phasing in of new technology. These committees receive advance notice
of impending technological changes, and when necessary. negotiate possible policies to mitigate the
negative effects with the collective bargaining unit. These policies incluc!ed advance notification to
workers. ayreenients to niin.irnize displacement, and provisions for retraining. Soine of the specific
clauses found in the contracts studied are listed below. A more detailed breakdown of clauses by by
union is shown in Table 14.
36 THE IMPACTS OF ROBOTICS
E a r n i n g s a n d Other-. C h a r a c t e r i s t i c s o f O r g a n i z e d W o r k e r s , May 1 9 8 0
U.S. Departirient o f Labor. Bureau o f Labor S t a t i s t i c s . September, 1981.
B t i l l e t i n 2105
37 THE IMPACTS OF ROBOTICS
e Transitional Allowances
e Severance Pay
e Retraining Provisions
These provisions have evolved over the years as part of an arrangement between the unions and
the firms to soften, or offset the impacts of displacement resulting from technological changes.
Technological change, in the view of the unions, results not only from the introduction of new labor
saving machinery, but also from design changes in the product, changes in engineering strategies,
and other types of modifications that "speed up the line", or reduce unit labor requirements. Another
intent of these provisions is to share some of the benefits of improved profitability with the workforce.
Provisions calling for the sharing of productivity benefits are based on the assumption that
technological improvements which increase productivity shduld in turn increase corporate profits. By
sharing the increased profits with the union, the company might improve the acceptance of new
technology. In the contracts studied, the UAW's Wage Improvement Factor was the only example of a
clause explicitly calling an annual percentage "productivity increase" exclushe of cost of living
increases. A UAW spokesman commented that this type of clause is only negotiated if the plant is in a
position to pay for it, and that where it has been negotiated, productivity has improved by more then
the wage improvement factor.
Paid Personal Holidays (PPH) are intended to spread fewer available jobs among a greater number
of employees by giving workers additional days off with pay in addition to holidays. The UAW has
negotiated twenty six Paid Personal Holidays over a three year period for each member working for an
automobile manufacturer.(About 50 percent of the UAW membership.) The intent is to reduce the
number of workers laid off by reducing the number of days worked per employee. Other unions have
implsmenterl similar plans b y increasing the standard vacation time. A UAW spokesman commented
that PPH's, like the Wage Improvement Factor, are negotiated when productivity is increasing within -
the p1:irit. and unit labor requirements are decreasing. The spokesinan also emphasized that PPH's
were only one part of a total package for offsetting displacement accompanying productivity
improvements. The additional paid holidays can also be viewed as another means of sharing the
benefits of increased productivity.
I
u u
-Io
O Y a u I .
6
I o 4 c u 0-
u
V x
U I Do
"
u
"
4 x
c-
0
0
u
0
V
.
U
5 U
L.
0 U
W
<
V
Y)
39 THE IPJPACTS 01' HOBO-TICS
Trnnsiiia/,n/ /1//owarices are provided to worii.eis when the firm transfers employees from plant-to-
plant. Thew allowances ranged from $500 to $1750 pa- employee in tht. four contracts we reviewed.
In some cases. benefits will also follow transferred employees Seniority does not transfer for the
IAM.
Advance Notice of technological change is required in all four of the contracts reviewed. The extent
of union input and involvement varied m i o n g the four unions. The UAlY and the IAM have committees
consisting of both union and management representation which would study and discuss each
change in tt?chnology, whereas, the IBEW contract indicates that management retains the sole right
of controlling the introduction of new technology.
Severance Pay is used to provide for workers who are permanently laid-off. It effectively pays
workers to leave their jobs. Severance pay is often used in cases of special retirement, where workers
are paid lump sums to leave the job, in addition to receiving a percentage of their original pension
benefits. This plan provides the firm with a quick, but costly means of reducing the size of the
workforce.
Three of the four unions studied have negotiated retraining provisions as'the responsibility of the
employer. The IUE has stated that it will make available specialized training for qualified workers
displaced by new technology. The UAW has training and retraining programs operating on an
ongoing basis.
The lntegriiy of the Bargaining Unit has also been negotiated in recent collective bargaining
agreements. The UAW has several agreements stating that all jobs previously in the bargaining unit
will stay in the unit. In other words, if an operator in a bargaining unit is replaced by a robot, then the
robot's operator will also be in the unit.
We are in the process of exploring the relationships between the increased use of robotics and the
potential for economic growth throughout the w h o l i economy. Even if robot users --primarily durable
goods produc=ers--wereto reduce their prodirction costs in real terms, we do not know if the rest of
the economy would experience high enough levels of economic growth to offset the predictable
losses in manufacturing' employment. The link between the main tangible benefit of robotics--
40 THE IMPACTS OF ROBOTICS
reduction in the cost cf capital- and the potential for economic growth throughout the whole
economy is still unexplored. The bottom /,ne is whether we can hope to realize a net social benefit--
including an net increase in employment-- by accelerating the use of robotics in manufacturing. It is
untf2rstood that there may be additional benefits in other areas, such as space or undersea
exploration. But it is important to know if the required levels of economic growth can be achieved in
the economy as it is now structured, without having to depend on the opening Lip of new frontiers. If
these growth and employment levels can not be achieved as a result of cost saving process
improvements in manufacturing, resources may have to be re'allocated to encourage the creating of
new products, services, and possibly, the development of new frontiers. This would require a
reevaluation of the current policy emphasis of stimulating economic growth by improving the
efficiency of creating "conventional" goods and services.
As mentioned earlier, we expect the primary quantifiable economic effect of robotics and
programmable automation to be a reduction in the real cost of manufacturing products made in small
to medium 'batches'--particularly, producers durable equipment. This rasies several important
questions. The first relates to how much of an impact robotics will have on the economics of batch
production. The second relates to the extent to which improvements in productivity in the capital
g'oods sector may impact the price. of output in other sectors that purchase these capital goods.
These linkages are shown in Figure 11.
A first step toward estimating the potential for productivity improvement in batch manufacturing is to
estimate the potential for reducing inventory carrying cost and set-up cost, and other benefits
associated with increased .lachine utilization However, the replacement of new robotic systems for
conventional technology may fundammtally alter the user's fixed and variable cost structure, as well
as create new technological and economic possibilities. Thus, the analysis is not so clear cut.
41 THE IMPACTS OF ROf30TICS
To trace the impacts of reductions in the rcal cost of capital goods on the price of all other goocls,
we can, in principli, identify the durable goods content of all other goods aild serviccs. For exainple,
even a piece of fruit has a large durable goods conteii-t since it requires planting, spraying, harvesting,
processing, plicking, and shipping. Input-output striictu.res, and capital flows matrices identify these
r&!tionships, and can ba used to estimate the extent to which the use of robotics might impact the
prices of the current bcrnd!e of goods in the economy. Unfortunately, there is not clear cut way to
estimate the extent to which a more productive manufacturing sector would spawn new economic
goods and services.
The limitations of trying to estimate the indirect impacts robotics and programmable automation
may have on econoinywide employment and econoinic growth should be acknowledged. While this
may be the curcial issue, it is also the hardest to analyze with any degree of precision. Eniployment
projections issued by the Office of'Economic Growth in the Department of Labor, and by other
research institutions (Chase, DRI, Wharton,etc.) estimate the growth of the labor force, based on
estimates of final aggregate demand. There are several problems with this procedure, primarily
imposed by the current limitations of economic science itself. One problem is that existing input -
output tables used in the analysis reflect historical--but not necessarily future-- technological
relationships. These models do not anticipate the basic structural changes we foresee in
maiiufacturicg, which might alter input- output relationships in important ways. Another comment is
that nwltiplier effects of productivity improvement are liot explicitly dealt with . These models do not
incorporate the feedback effects these chanjes might induce in other sectors of the economy.
Another important restriction is the lack of knowledge about price and demand relationships,
particularly about how changes in price might trigger substitution effects. While the BLS, and other
forecasters have already published projections of economy wide employment for the end of this
decade, it must be pointed out that these models, in their current form, are not intended to assess,
and may not even be capable of assessing the impacts of robotics and programmable automation on
ernployment.either in manufacturing, or throughout the whole economy.
Most of the published literature on robots describes physical capabilities and particular
applications, or deals with the narrowly defined economics of robot use, based primarily on the
difference between amortized robot cost and the "all-included'' cost of hourly labor. Discussions of
human factors, if any, tend to be sweeping statements about the importance of gaining the
acceptance of workers and top management support, limiting human factors concerns'to bypassing,
or eliminating potential pockets of resistance to robotics. There has been little serious discussion to
date of how to cope with the hard reality of developirlg needed new work skills on the one hand and
how to deal with people who have obsolescent skills, on the other.
Robot iisers have been reluctant to discuss plans for robot use in the future, even though many
manufacturers are testing applications. They argue that such information must be kept confidential
for competitive reasons. One result of private industry's uncommunicative attitude about future plans
is that very little is being done to warn or prepare those workers whose jobs may be eliminated, or
sirbstantially modified as a direct, or indirect result of introducing robots. In the absence of solid facts,
or even informed speculations as to what types of adjustments might occur, and their tims phasing
42 THE IMPACTS OF ROBOTICS
2nd mnynitude, unions, media reporters, and governrwnt officials have Starit?d to suspt'ct the worst,
and acjk: How many people will lose their jobs as this new wave of automation sweeps through
industry ? Private inctustry undoubtedly has an interest in the pi!blic perception of the impacts of
robots GI? the labor force. If the phasing in of robots is handlecl ineptly and insensitively,( or if people
evzn think this is the case), unions, and other factions of society might conceivably find enough
common interest-- based on a fear of technology-- to organize a "Neo Luddite" attack on robots and
other forms of automation. Short of this extreme scenario, widespread social dissension could occur,
fed by distrust of business and dissatisfaction with the record of a capitelist society in dealing with
fester iny soc ia1 probIe111 s.
To develop the necessary human capital at both the institutional and individual level, and to smooth
the short term transitory impacts on the labor force, all the major actors must cornmit themselves to a
cooperative effort to prepare and assist the workers most likely to be affected by the changes to
come. To effectively prevent social trauma due to rapid introduction of robotics, without impeding
technological progress itself requires:
0 Long Range Planning by industry and government for future employment needs and new
job skill requirements.
e The provision of effective education and training facilities to upgrade workers frbm skill
categories that are, or will be in surplus supply to skill categories that are scarce.
0 The provision of effective facilities to locate suitable jobs and place workers in them,
with relocation assistance if necessary.
It is in industry's interest to assume a more active role in planning future employment needs. It
must ensure that the workforce gets an accurate preview of the requirements of tomorrow's
workplace, and that the appropriate skills are sought and taught.
Colleges and universities in the United States do a reasonably good job at educating science and
traditional engineering students. But many of our existing educational institutions do not have the
capability. or even the inclination to involve themselves in training unskilled OF semi-skilled people for
operational functions in indugtry. Experience with publically sponsored training programs suggest
that. while they are reasonably capable of retraining skilled workers to do new jobs, they have
seldornly been successful at training the "hard core" unemployed to be productive. The educational
establishment must face up to this problem , since some of the factory jobs which have historically
en?a:oyed the least skilled workers--such as material handlers and machine loaders/unloaders-- will
eventiinlly be replaced by robots. The same Is true of many semi-skilledjobs such as welding.
The educational establishment also needs to face up to several important deficiencies. There is too
much emphasis on education for white collar jobs as opposed to training people to work with (i.e.
sclp?rvise. inaintain, and repair) machines. It appears that blue collar and skilled workers do not have
a favorable irnage in our society, despite the fact that many of these jobs require more schooling, and
pay coinpnrable wages. Trade school is often viewed as an alternative for students who flunk out of
43 THE IMPACTS OF ROBOTICS
the academic track, or for cit~linquents. The more capable students are steering away froin factory
viork.
The Unions and management need ways of interacting cooperatively--rather than BS adversaries-
for dealing with issues of displacement and of changes in the workplace.
It would seem that if industry continues its uncommunicative policy , the unions will continue to
emphasize setting precedents in order to ensure their survival in an uncertain environment. This type
of "gaming " obstructs th.e type of planning that both unions and management need to do in
cooperation with each other to solve teal problems and achieve mutual benefits. It is not reasonable
to expect firms to be more open with the unions if such disclosures would would constrain thein in
what type of technology they could develop, or how they could use it. Neither is it reasonable to
expect unions to be more cooperative with management, and more flexible in their bargaining
positions, if such an attitude would threaten the security of their workers and the long term viability of
the unions themselves. The only way for both sides to break out of this bind is for government to
change the conditions under which unions and industry talk to each other. In this context, the US may
have much to !earn from Japan, Germany, and other industrial countries.
Another of the government's key roles should be to provide incentives which would induce industry
to take positiv2 action on upgrading its human resources now. For example, the government could
give tax incentives to partially reimburse industry for educatioh and training investments in their
employees. It could provide more favorable tax treatment for individuals who undertake formal
retraining programs in mid-career. And, of course, it could provide inducements (financid and other)
to educational institutions to induce them to redirect their efforts into new areas.
Education and training are established functions of all levels of government. It is vital that
publically funded education/training programs reflect the emerging-- rather than the obsolete - -
needs of industry and society . . Vocational education enrollments and completions in six
metalworking occupational categories are shown in Table 15. These six categories account for just
over three percent of all vocational education enrollments for FY 1978.'5 Several popular
occupational categories for publically funded training programs are precisely those which have been
identified as prime candidates for robots. It appears that public education institutions in the U S have
not yet recognized the future employment skill needs of society. Training programs funded directly by
government have an incentive to get people through a program quickly, and document their
"success", even if they are providing people with obsolescent skills.
The future outlook for .employment within most factory occupations cannot be extrapolated from
historical data. The basic technological relationships governing the mix .of labor and capital required
to satisfy a given level of output are changing in very fundaniental ways. Yet , government
publications are still projecting employnient requirements for many of the factory occupations without
any explicit acknowledgement the impact of emerging production technologies--iricluding robotics.
Long range planning of .:mployment requirements and identification of vulnerable job categories
15Mach~nestand machlne operatrvrs. and welders account for slightly less then foul percent 01 t h e employed workforce
44 THE IMPACTS OF ROBOTICS
Source:
O c c u p a t i o n a l P r o j e c t i o n s and T r a i n i n g D a t a , 1980 e d i t i o n
U.S. Department o f Labor
B u r e a u o f L a b o r S t a t i s t i c s . B u l l e t i n 2052
45 THE IMPACTS OF ROBOTICS
cannot be carried otit b y governinent aymcics, such as the Bureau of Labor Statistics, without inputs
frorii industry. N a t t i e r iiidusiry nor society at large cai) afford th? c ~ n s e q u c n c e sof having too many
people steered into obsolescent occupations while there are too few people with badly needed skills.
46 THE IMPACTS OF ROBOTICS
KEY
P # = patent number
A: Date patent applied for
I: Date patent issued
Rcfc rcnces
[Albus 811 James S. Albus.
Industrial Robots and Productivity lmpro ven?en t .
Technicat Report, lncfustrial Systems Division, National Bureau of Standards,
October, 1981.
Draft presented at the Office of Technology Assessment Robotics Workshop, July
31, Washington, D.C.
P~P
791P Neale W. Clapp.
Three Laws for Robotocist: An Approach to Overcoming Worker and Management
Resistance to Industrial Robots.
Technical Report tvlS79-775, Society of Manufacturing Engineers, March, 1979.
[CMU 811
The Impacts ofRobotics on the Workforce or Workplace.
Technical Report, Carnegie Mellon University, June, 1981.
A student project cosponsored by the Department of Engineering and Public
Policy, the Department of Social Science, the School of Urban and Public
Affairs.
1